To evaluate a commercialized in situ hybridization (ISH) assay for detecting human papillomavirus (HPV) DNA, we compared the ability of a new ISH probe, Inform HPV III (Ventana Medical Systems, Tucson, AZ), to that of PCR assays to detect HPV DNA in cervical tissue specimens with normal cervix (20 cases), cervical intraepithelial neoplasia (CIN; CIN 1, 27 cases; CIN 2, 28 cases; and CIN 3, 33 cases), and cervical carcinoma (29 cases). General HPV DNA was detected using consensus primer-mediated PCR assays. HPV genotyping was performed by using EasyChip HPV blot (King Car Yuan Shan Institute, I-Lan, Taiwan). HPV16 integration status (E2/E6 ratio) was determined by using quantitative real-time PCR. Our findings showed that the ISH and PCR had fair to good agreements in detecting HPV DNA across all CIN categories without significant differences (Kappa coefficient, 0.34 to 0.63; P = 0.13 to 1.0). However, ISH detected significantly fewer HPV-positive cases in carcinoma than PCR did (Kappa coefficient, 0.2; P = 0.03). Eleven cases with ISH- PCR+ results had HPV types that can be detected by Inform HPV III. Five carcinoma cases with ISH- PCR+ results showed a significantly higher level of integrated HPV16 (P = 0.008) than did the ISH+ cases. As a consequence, lower copy numbers of episomal HPV16 in carcinoma might be the cause for the false-negative ISH results. Although the punctate signal pattern of HPV significantly increased with the severity of disease (P trend = 0.01), no significant difference in the HPV16 integration status was observed between the cases with a punctate signal only and the cases with mixed punctate and diffuse signals (P = 0.4). In conclusion, ISH using the Inform HPV III probe seems comparable to PCR for detecting HPV DNA in cervical tissue with CINs. False-negative ISH results appear to be associated with the lower copy numbers of the episomal HPV16 but not with the ability of the Inform HPV III probe to detect specific HPV types. In addition, signal patterns, especially a mixed punctate and diffuse pattern of HPV, cannot be reliably used to predict viral integration status.